Taiwan Association of Engineering and Technology Innovation: E-Journals
Not a member yet
887 research outputs found
Sort by
Hardware Implementation of Chaotic Image Encryption on FPGA
Full text linkThis paper proposes a new FPGA-based architecture for secure image encryption, utilizing chaotic maps for permutation and substitution mechanisms. Specifically, Duffing map is employed to generate permutation addresses, and the Henon map for value substitution; both techniques produce pseudo-random sequences with strong sensitivity to initial conditions. The hardware design implemented using Xilinx System Generator and deployed on an Artix-7 FPGA supports both grayscale and RGB image formats. A comprehensive performance analysis, utilizing Mean Squared Error, Histogram analysis, Correlation Coefficient Analysis, Number of Pixels Change Rate, and Unified Average Changing Intensity, demonstrates perfect image recovery and robust encryption resistance. The keys used in the encryption system are passed using the NIST STS randomness test. The single-round and multi-round designs deliver a processing acceleration ranging from 2.11 to 3.92 for image sizes 128×128 to 1024×1024 pixels, highlighting their effectiveness and practicality for real-time encryption in low-latency environments
A Novel Method for Detecting Voltage Anomaly in Distribution Networks Based on Improved Standard Deviation Filters
Full text linkAccurate voltage anomaly detection is the prerequisite for the reliable operation of the distribution network. However, the traditional detection methods are prone to missed and false alarms. In practice, the distribution of phase voltage difference satisfies a normal distribution during normal operation and deviates from the distribution during faults. This paper proposes a novel method for voltage anomaly detection in distribution networks based on an improved standard deviation filter. The proposed method identifies an anomaly by evaluating the dispersion degree based on the mean and standard deviation of the phase voltage difference dataset. The short-term cycle, long-term cycle, and weighting coefficient are adopted rationally, thus large data storage requirements and repeated calculations can be avoided. Compared with the clustering and the isolation forest algorithms, the proposed method can identify voltage anomalies more accurately. The reliability of the proposed method is verified by on-site data
Information Security Protection System for Networked OT Environments of Industrial Control in Smart Manufacturing
Full text linkThis study develops an innovative information security protection system for end devices in smart manufacturing industrial control environments. By employing six key functionalities—lightweight identity authentication, traffic analysis, key management, personnel authorization control, system status monitoring, and an alarm mechanism—the system addresses the limitations of traditional firewalls. Experimental procedures involved testing the system against common threats, including phishing (fraud), physical intrusion, and Denial of Service attacks. Results demonstrate over 90% success in mitigating these attacks while maintaining operational efficiency. Furthermore, real-time monitoring and alert features enhance data protection and ensure reliable factory operations
Enhancing Learning Outcomes in Mechanical Drawing Practice Courses Using Student Teams Achievement Divisions
Full text linkThis study applies the STAD teaching method in a mechanical drawing course for 14 tenth-grade vocational students in Taiwan. It uses a three-cycle action research design to verify its effectiveness empirically. The results show improved learning outcomes, notable progress among female students, and enhanced self-efficacy. However, STAD slightly reduced students’ willingness to engage in peer discussions. In addition, the method supports teacher professional development. Future research is recommended to explore its broader applicability and long-term effects
FCA-ResNet: An Improved Model with Enhanced Multi-Scale Feature Fusion and Coordinate Attention for Wheat Leaf Disease Classification
Full text linkRapid and accurate identification of leaf disease is essential in intelligent agriculture. Current methods often struggle with balancing precision and speed. This research introduces the fusion coordinate attention and residual network (FCA-ResNet) model to improve classification accuracy while maintaining a lightweight structure for both healthy wheat leaves and five common wheat leaf diseases. FCA-ResNet incorporates a coordinate attention (CA) mechanism along with a multi-branch Inception module. The model consists of an Inception-based multi-branch structure and CA mechanism fusion module, which optimizes feature focus and weight allocation. Additionally, a multi-scale fusion module utilizes both channel and spatial attention mechanisms to effectively integrate shallow and deep features, improving the detection accuracy of small lesions. The multi-branch structure is designed to replace traditional multi-layer convolution, resulting in a lightweight model. The model achieves an average accuracy of 91.6% on custom datasets, demonstrating its effectiveness in plant disease detection for agriculture
Iterative Clipping Filtering and Saleh Model Predistorter on a MIMO-OFDM System Testbed Using Software Defined Radio
Full text linkThe purpose of this study is to address the challenges of high peak-to-average power ratio (PAPR) and nonlinear power amplifier (PA) distortion in multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems based on the IEEE 802.11ac. The research integrates iterative clipping filtering (ICF) for PAPR reduction and the Saleh model predistortion (PD) for PA linearization. Implemented on a software defined radio (SDR) platform using NI-USRP devices, the system is evaluated in real-world line-of-sight (LOS) and non-line-of-sight (NLOS) environments. Results show significant PAPR reduction, from 19.7 dB to 10.3 dB, and improved PA linearity, achieving a 94.80% error vector magnitude (EVM) reduction. Furthermore, the combined approach exhibits lower symbol error rates (SER) and error-free data transmission, particularly under LOS conditions. Compared to conventional methods, the system demonstrates superior execution efficiency with 475–503 ms processing times
Optimizing Nanofluid Minimum Quantity Lubrication Machining of Inconel-800 Using Kriging Non-Dominated Sorting Genetic Algorithm II
Full text linkThis study optimizes the machining process of Inconel-800 superalloy using nanofluid minimum quantity lubrication (MQL) with multi-wall carbon nanotubes (MWCNTs) and biodegradable coconut oil. A Taguchi design with 27 trials is used to examine the effects of varying nanoparticle concentrations and machining parameters on surface roughness and temperature. The optimized nanofluid MQL system improves surface roughness by 26.22%, reduces surface roughness peak-to-valley by 12.06%, and significantly lowers temperature, demonstrating improved quality and thermal management. A Kriging model predicts outcomes with high accuracy (R2 > 0.9), and multi-objective optimization using Kriging and the non-dominated sorting genetic algorithm II identifies an optimal balance between surface roughness and temperature. Additionally, using coconut oil as the lubricant base in the nanofluid MQL system promotes sustainable machining by reducing reliance on conventional lubricants and environmental impact. These findings validate the effectiveness of advanced optimization techniques combined with nanofluid MQL for superior sustainable machining of superalloys
Energy Demand Forecasting for Hybrid Microgrid Systems Using Machine Learning Models
Full text linkThis study aims to design energy demand forecasting models for energy management in hybrid microgrid systems using optimized machine learning techniques. By incorporating temperature, humidity, season, hour of the day, and irradiance, the complex relationship between these input parameters and the yield of photovoltaics, generator, and grid energy sources is examined. Five different machine learning models including linear regression, random forest (RF), support vector regression, artificial neural network, and extreme gradient boosting models are adopted in this study. Evaluation of model performance shows that the RF model is the best candidate for the dataset, with a mean-squared error of 0.2023, mean absolute error of 0.0831, root-mean-squared error of 0.4498, and R² score of 0.9992. Shapley additive explanations analysis identified key predictors such as hour, irradiation, and season while highlighting the negative impact of humidity and day of the week on energy demand
A Novel Data Transmission Model Using Hybrid Encryption Scheme for Preserving Data Integrity
Full text linkThe objective of the study is to introduce a novel hybrid encryption scheme, combining both symmetric and asymmetric encryptions with a data shuffling mechanism, to enhance data obfuscation and encryption security. The approach uses RSA for asymmetric encryption and ChaCha20-Poly1305 for symmetric encryption. To increase the complexity, an additional phase involves reorganizing the RSA-encrypted data blocks. Furthermore, symmetric key generation using the key derivation function is employed to generate the key for symmetric encryption through an asymmetric private key. Decryption entails reversing these procedures. This model significantly enhances security through an additional shuffling step, measured by performance metrics like encryption and decryption times, throughput rate, and the avalanche effect. The method, despite increasing execution time compared to symmetric models, yields comparable results for asymmetric models and ensures robustness. The proposed method outperforms traditional methods regarding resistance to cryptanalytic attacks, including chosen-plaintext and pattern analysis attacks
Feasibility of 3D Printing on Environmentally Friendly Cementless Materials with Low Thermal Conductivity
Full text linkThis study investigates ultra-fine fly ash (UFA) and co-fired fly ash (CFA) to produce binary cementless binders without alkali activators and determines the effects of molding temperatures (17 ℃, 50 ℃, 60 ℃, 70 ℃, 80 ℃, and 90 ℃) on thermal conductivity and microstructures. The pastes are subjected to flow and expansion tests to verify the mixing state of the two industrial by-products for a fixed water-to-binder ratio of 0.4. Compressive strength, water absorption, density, thermal conductivity, and scanning electron microscope analyses determine material properties and the optimal molding temperature. Results reveal that higher hardening temperatures lead to higher water absorption and lower density. The 50 ℃ specimen exhibits the lowest thermal conductivity of 0.1796 W/m·K at 56 days. The printed specimens with UFA and CFA at a 1:1 ratio achieve a 28-day compressive strength of 9 MPa and a thermal conductivity of 0.2064 W/m·K